It’s a PIC and they tend to handle a lot of different crystals, resonators, and external oscillators, but you occasionally have to change the programming to match what you’re using. The CV-12 spec just says “a 16MHz crystal and 2 x 22 pF capacitors” so seems they’ve designed for a load capacitance of ~15 pF. If you have something else, you should be able to tweak the caps.
(rule of thumb is to use 2×(CL−5) pF where CL is the load capacitance and 5 pF is estimated trace capacitance)
nearly went off to buy some expensive 4504’s good job I re-read the first post, and looked in the packet!
Yes, I must say I’m impressed that Sam has been able to fulfill his promise to supply some of the more critical components with some of the kits.
Wanted to start building this, but I noticed that I don’t have the 1uF Tantalum cap… why is tantalum used here? Can I get away with electrolytic? 
Not entirely clear why that’s specified as a 1 uF tantalum, or for that matter why it’s a 1 uF. The original LM317 circuit had a 1 uF there for transient response iiuc (it doesn’t really need one for stability), but the TL431 gets less stable if you use it with the wrong capacitive loads:
iirc the current through it is low (1 mA?) so the unsafe range would be maybe 20 nF to 1.5 uF (just outside curve C near 0 mA in that diagram) and their application note suggests going with 10× that to make sure you have plenty of design margin. So a 10 uF electrolytic should be good, maybe?
@fredrik the tantalum EDIT I just re read your writing and removed the last post to avoid confusion, so yes, I put it back in as I did not question it. it works and is fine. but I can imagine like fredrik says a normal electrolytic is fine, please let us know your findings @Sebastian as I may update!!!
In the last prototype I learnt the need for the cap! as I just took it out. definitely needed. as without a cap there the notes get awfully unreliable! so definitely required for stability of the 4504’s. but give a 10uf a go ill be very interested.
that being said for anyone on the 1uf tantalum already, if you see on fredrik graph its right on the cusp of stable, and I can confirm it is stable! so it’s still fine.
edit, ill give just a standard 10uf a go and report back. As it would definitely be a pro! if that one does an adequate job, it simplifies the Bill of Materials somewhat
Okai, great, I will also try a 10uF EL then!
Another thing: R26 is listed as 270 ohm on the BOM, but on the pcb it says 220 – 270. So, 270 is better, but 220 could work? Bit confused again 
It depends on your supply voltage – if it’s good enough to give you ~11.5 V or above, you can use 270 ohm and save a tiny bit of power. If it may drop below that, you may have to use a lower value to make sure the TL431 gets the 1 mA it needs to regulate. See discussion in this thread:
EDIT: 220 ohm is probably better, see here.
Okai, but is there a disadvantage of using 220 ohm, apart from a saving a bit of power? Sounds like it should be negligible…
On another note: I assume there is no problem if I use a “regular” (TO220) TL7805 instead of the TO92? (I need to check the pinout though, and check if the larger legs fit…)
As always: Thank you all!
As I was assembling the circuit, I didn’t have the 5.1k resistor specified for R4 in series with the RV1 trim pot, so I went to revisit our calculations to see if a more common 4.7k would do, and I noticed that we might have forgotten to figure in the approx. 1mA flowing through R4 and RV1 in our computations.
So we might not have that “1mA to spare” after all.
I would recommend going with 220 ohms for R26
It gives us (11.5-10.667)/220 = 3.8mA which is 1mA for the TL431, 1.3mA for the circuit, about 1mA for R4&RV1, and 0.5mA to spare.
I’ll be using 220 ohms for R26 and 4.7k for R4
Hm… I have 220 ohms for R26, the EL instead of tantalum and 5.1k for R4. But I can not get the voltage test point higher than 9.9V… I can go as low as 5V though xD. My 12V Rail has ~13V, so it should work… I will check more tomorrow, too tired now 
Reflowed the parts and now the voltage is fine!! But I noticed that I got 6N 138 instead of 137 for midi optocouplers :-/ need to wait for the next order again xD
Hm… all gate LEDs are always on… :-/ what’s going on?
You mentioned having 13V on your +12V rail earlier, that is really not normal, so maybe you have a bad (blown?) power supply and have damaged either the micro-controller or the buffer opamps.
Is the +12V back to closer to 12V? How’s your -12V?
Are the CV outputs working normally? At all?
Rails are +12.3V and -11.8V. It probably was like this before as well. CV is always 0V, but keep in mind that I have the wrong optocoupler and hence have not midi input to test… maybe the problems go away as soon as I have the correct one, but I think all the gates being high is still strange… I will measure the micro-controller pins directly next…
Hm… using the correct 6N137 did not help :-/ When I pull out the microcontroller all LEDs are off. The CV outputs are all 0V with the microcontroller connected and some midi coming in. Any ideas how I can test the chip?
Maybe something with the crystal? I got this one, should be correct, right?
Edit:
When I measure pin4 of the chip (atmega), I measure ~20mV, but when I measure the voltage after the opamp, I measure 5V at the output jack! how can that be? some opamp magic?
(to follow in the schematic: Pin4 goes to label 12A, which goes to 12B via the interboard connectors and then to the non inverted input of U11A. U11A is wired as a voltage buffer and its output goes via 100ohm to the output jack)
have you connected the chain input together??
also if you have a picture of the back it may help!
